Index: /CKEditor/branches/features/pasting/_source/core/dtd.js =================================================================== --- /CKEditor/branches/features/pasting/_source/core/dtd.js (revision 4279) +++ /CKEditor/branches/features/pasting/_source/core/dtd.js (revision 4280) @@ -54,5 +54,6 @@ Q = {li:1}; - var block = {address:1,blockquote:1,center:1,dir:1,div:1,dl:1,fieldset:1,form:1,h1:1,h2:1,h3:1,h4:1,h5:1,h6:1,hr:1,isindex:1,menu:1,noframes:1,ol:1,p:1,pre:1,table:1,ul:1}; + var block = {address:1,blockquote:1,center:1,dir:1,div:1,dl:1,fieldset:1,form:1,h1:1,h2:1,h3:1,h4:1,h5:1,h6:1,hr:1,isindex:1,menu:1,noframes:1,ol:1,p:1,pre:1,table:1,ul:1}, + nonEmptyInline = { abbr:1,acronym:1,b:1,bdo:1,big:1,cite:1,code:1,del:1,dfn:1,em:1,font:1,i:1,ins:1,label:1,kbd:1,q:1,samp:1,small:1,span:1,strike:1,strong:1,sub:1,sup:1,tt:1,u:1,'var':1 }; return /** @lends CKEDITOR.dtd */ { @@ -66,4 +67,11 @@ */ $block : block, + + /** + * List of non-empty inline elements, like "b" or "a". + * @type Object + * @example + */ + $nonEmptyInline : nonEmptyInline, $body : X({script:1}, block), Index: /CKEditor/branches/features/pasting/_source/plugins/htmldataprocessor/plugin.js =================================================================== --- /CKEditor/branches/features/pasting/_source/plugins/htmldataprocessor/plugin.js (revision 4279) +++ /CKEditor/branches/features/pasting/_source/plugins/htmldataprocessor/plugin.js (revision 4280) @@ -78,5 +78,7 @@ // Event attributes (onXYZ) must not be directly set. They can become // active in the editing area (IE|WebKit). - [ ( /^on/ ), '_cke_pa_on' ] + [ ( /^on/ ), '_cke_pa_on' ], + // Bogus attributes. + [ /^cke:/, '' ] ], elements : Index: /CKEditor/branches/features/pasting/_source/plugins/pastefromword/plugin.js =================================================================== --- /CKEditor/branches/features/pasting/_source/plugins/pastefromword/plugin.js (revision 4279) +++ /CKEditor/branches/features/pasting/_source/plugins/pastefromword/plugin.js (revision 4280) @@ -50,4 +50,9 @@ CKEDITOR.plugins.pastefromword = { + _ : + { + cachedRules : null + }, + utils : { @@ -106,4 +111,11 @@ if( /mso-list:\s*Ignore/i.test( styleText ) ) return true; + }, + + isContainingOnlySpaces : function( element ) + { + var text; + return ( ( text = element.onlyChild() ) + && /^(:?\s| )+$/.exec( text.value ) ); }, @@ -142,5 +154,6 @@ }, - filters : { + filters : + { /** * A simple filter which always rejecting. @@ -258,5 +271,11 @@ if( element.name.indexOf( 'cke:' ) == -1 ) return false; - } + }, + + /** + * A filter which will be used to apply inline css style according the stylesheet + * definition rules, is generated lazily when filtering. + */ + applyStyleFilter : null }, @@ -264,5 +283,11 @@ getRules : function( editor ) { - var filters = this.filters, + var cached; + if( cached = this._.cachedRules ) + return cached; + + var dtd = CKEDITOR.dtd, + listDtdParents = dtd.parentOf( 'ol' ), + filters = this.filters, falsyFilter = filters.falsyFilter, stylesFilter = filters.stylesFilter, @@ -272,6 +297,6 @@ createListBulletMarker = this.utils.createListBulletMarker, isListBulletIndicator = this.utils.isListBulletIndicator, - resolveList = this.utils.resolveList, - listDtdParents = CKEDITOR.dtd.parentOf( 'ol' ), + containsNothingButSpaces = this.utils.isContainingOnlySpaces, + resolveListItem = this.utils.resolveList, config = editor.config, ignoreFontFace = config.pasteFromWordIgnoreFontFace; @@ -291,30 +316,35 @@ var tagName = element.name || ''; - // Resolve inline css style for Firefox. + // Firefox: adding inline style. var applyStyleFilter; if( CKEDITOR.env.gecko && ( applyStyleFilter = filters.applyStyleFilter ) ) applyStyleFilter( element ); - + // IE leave empty spaces at the beginning of body. + if ( !tagName ) + { + var textNode = element.firstTextChild(); + if ( textNode && textNode.value.match( /^(:?\s| )+$/ ) ) + element.children.splice( 0, 1 ); + } // Processing headings. - if ( tagName.match( /h\d/ ) ) + else if ( tagName.match( /h\d/ ) ) { element.filterChildren(); - // Heading might be a list. - if( resolveList( element ) ) + // Is the heading actually a list item? + if( resolveListItem( element ) ) return; - // Migrate heading formatting to editor configured ones. + // Adapt heading styles to editor's convention. elementMigrateFilter( config[ 'format_' + tagName ] )( element ); } // Remove inline elements which contain only empty spaces. - else if( tagName.match( /^(:?b|u|i|strike|span)$/ ) ) + else if( tagName in dtd.$nonEmptyInline ) { element.filterChildren(); - var child = element.onlyChild(); - if ( child && /^(:?\s| )+$/.exec( child.value ) ) + if( containsNothingButSpaces( element ) ) delete element.name; } - // Remove ms-office namespaced element while preserving the content. + // Remove ms-office namespaced element, with it's content preserved. else if( tagName.indexOf( ':' ) != -1 && tagName.indexOf( 'cke' ) == -1 ) @@ -322,5 +352,5 @@ element.filterChildren(); - // Restore img element from vml. + // Restore image real link from vml. if( tagName == 'v:imagedata' ) { @@ -329,23 +359,11 @@ element.attributes.src = href; element.name = 'img'; + return; } - else - delete element.name; - } - else if ( !tagName ) - { - // Trim empty spaces at the beginning of document for IE. - if( CKEDITOR.env.ie ) - { - var firstTextChild = element.firstTextChild(); - if ( firstTextChild - && firstTextChild.value.match( /^(:?\s| )+$/ ) ) - element.children.splice( 0, 1 ); - } - } - - // Any dtd-valid element which could contain a list. - if( !tagName && element.children - || tagName in listDtdParents ) + delete element.name; + } + + // Assembling list items into a whole list. + if( !tagName || tagName in listDtdParents ) { element.filterChildren(); @@ -411,12 +429,8 @@ list = null; } - - // Filter childrens again for cleaning up - // the list attributes. - element.filterChildren(); } }, - // We would like drop any style sheet, but Firefox conclude - // certain styles in it, so we're required to change them into + // We'll drop any style sheet, but Firefox conclude + // certain styles in it, so they're required to be changed into // inline ones. 'style' : function( element ) @@ -424,26 +438,29 @@ if( CKEDITOR.env.gecko && !filters.applyStyleFilter ) { + // Grab only the style definition section. var styleDefSection = element.onlyChild().value.match( /\/\* Style Definitions \*\/([\s\S]*?)\/\*/ ), styleDefText = styleDefSection && styleDefSection[ 1 ], - rules = {}; + rules = {}; // Storing the parsed result. if( styleDefText ) { styleDefText.replace(/[\n\r]/g,'') // remove line-breaks. - // Extract selectors and rules. - .replace( /(.+?)\s*\{(.+?)\}/g, + // Extract selectors and style properties. + .replace( /(.+?)\{(.+?)\}/g, function( rule, selectors, styleBlock ) { selectors = selectors.split( ',' ); - var length = selectors.length, - selector; + var length = selectors.length, selector; for ( var i = 0; i < length; i++ ) { + // Assume MS-Word mostly generate only simple + // selector( [Type selector][Class selector]). CKEDITOR.tools.trim( selectors[ i ] ) - .replace( /^(\w+?)(\.[\w-]+)?$/g, + .replace( /^(\w+)(\.[\w-]+)?$/g, function( match, tagName, className ) { tagName = tagName || '*'; className = className.substring( 1, className.length ); + if( !rules[ tagName ] ) rules[ tagName ] = {}; @@ -479,26 +496,26 @@ var attrs = element.attributes, parent = element.parent, - children = element.children, - firstChild = children && children[ 0 ]; + children = element.children; // Drop the single wrapper paragraph within table cell - // while preserve the styles. + // with all attributes preserved to cell. if( /td|th/.test( parent.name ) && parent.onlyChild() ) { - if( attrs && attrs.style ) - parent.addStyle( attrs.style ); - + CKEDITOR.tools.extend( parent.attributes, attrs ); + attrs && parent.addStyle( attrs.style ); delete element.name; return; } - // Paragraph might be a list. - if( resolveList( element ) ) + // Is the paragraph actually a list item? + if( resolveListItem( element ) ) return; - // Migrate paragraph formatting based on editor's enter-mode. + // Adapt paragraph formatting to editor's convention + // according to enter-mode. if( config.enterMode == CKEDITOR.ENTER_BR ) { + // We suffer from attribute/style lost in this situation. delete element.name; element.add( new CKEDITOR.htmlParser.element( 'br' ) ); @@ -519,18 +536,21 @@ element.filterChildren(); - // Merge nested tags. - var parent = element.parent; - if( 'font' == parent.name ) + + var attrs = element.attributes, + parent = element.parent; + + if( 'font' == parent.name ) // Merge nested tags. { CKEDITOR.tools.extend( parent.attributes, element.attributes ); + attrs && parent.addStyle( attrs.style ); delete element.name; return; } - // Convert the topmost into a span with font-styles. + // Convert the merged into a span with all attributes preserved. else { - var attrs = element.attributes, - styleText = ''; + var styleText = attrs.style || ''; + // IE's having those deprecated attributes, normalize them. if ( attrs.color ) { @@ -552,8 +572,6 @@ delete attrs.size; } - element.name = 'span'; element.addStyle( styleText ); - } }, @@ -574,10 +592,11 @@ // Update the src attribute of image element with href. var children = element.children, - firstChild = children && children[ 0 ], - secondChild; - if( firstChild - && 'cke:imagesource' == firstChild.name ) - { - secondChild = children[ 1 ]; + attrs = element.attributes, + styleText = attrs && attrs.style, + firstChild = children && children[ 0 ]; + + if( firstChild && firstChild.name == 'cke:imagesource' ) + { + var secondChild = children[ 1 ]; if ( 'img' == secondChild.name ) secondChild.attributes.src = firstChild.attributes.src; @@ -586,9 +605,8 @@ } - // Migrate font related styles. - var attrs = element.attributes, - styleText = attrs && attrs.style; + // Assume MS-Word mostly carry font related styles on , + // adapting them to editor's convention. if( styleText ) - attrs.style = stylesFilter( + attrs.style = stylesFilter( [ [ /^font-family$/, null, styleMigrateFilter( config[ 'font_style' ], 'family' ) ], @@ -606,13 +624,6 @@ 'sup' : elementMigrateFilter( config[ 'coreStyles_superscript' ] ), 'sub' : elementMigrateFilter( config[ 'coreStyles_subscript' ] ), - 'v:imagedata' : function( element ) - { - var href = element.attributes['o:href']; - if( href ) - element.attributes.src = href; - element.name = 'img'; - }, - // Editor doesn't support anchor with content currently(#3582), - // drop such anchors. + // Editor doesn't support anchor with content currently (#3582), + // drop such anchors with content preserved. 'a' : function( element ) { @@ -636,8 +647,4 @@ attributes : { - 'cke:listtype' : bogusAttrFilter, - 'cke:indent' : bogusAttrFilter, - // Remove mso-xxx styles. - // Remove margin styles. 'style' : stylesFilter( [ @@ -675,16 +682,18 @@ ignoreFontFace ? [ /font-?/ ] : null, ] ), + // Prefer width styles over 'width' attributes. 'width' : function( value, element ) { - // Prefer width style over attribute on table cell. - if( element.name in { td : 1, th : 1 } ) + if( element.name in dtd.$tableContent ) return false; }, - - 'class' : function( value ) - { - if( value.match( /^(:?Mso|Spell|ListParagraph)/i ) ) + // Prefer border styles over table 'border' attributes. + 'border' : function( value, element ) + { + if( element.name in dtd.$tableContent ) return false; }, + + 'class' : falsyFilter, // MS-Word always generate both 'text-align' along with @@ -693,19 +702,13 @@ 'align' : falsyFilter, 'bgcolor' : falsyFilter, + // Deprecate 'valign' attribute in favor of 'vertical-align'. 'valign' : function( value, element ) { - // TODO: The style chang doesn't work now because of filtering system. if( value != 'top' ) element.addStyle( 'vertical-align', value ); return false; - }, - - // Avoid table 'border' attribute in favor of cell border styles. - 'border' : function( value, element ) - { - if( element.name == 'table' ) - return false; } + }, @@ -726,9 +729,8 @@ return new CKEDITOR.htmlParser.element( 'cke:imagesource', { src : imageSource[ 1 ] } ); // Seek for list bullet style indicator. - else if ( listInfo ) + if ( listInfo ) { var listSymbol = listInfo[ 1 ], listType = listSymbol.match( />([^\s])([.)]?)